A professional-grade speech recognition skill leveraging Alibaba's FunASR model via the DashScope API for high-accuracy audio transcription.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install fun-asr
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install fun-asr using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Fun-ASR?
Fun-ASR is a sophisticated end-to-end speech recognition large model developed by the Alibaba DAMO Academy team. It is trained on tens of millions of hours of real-world speech data, providing the model with exceptional contextual understanding and industry adaptability. By integrating this as part of your Openclaw Skills library, developers can enable AI agents to process complex audio files with high precision.
This skill interfaces directly with the Alibaba DashScope API to transform spoken language into structured text. It is designed for developers who require robust, scalable, and commercially viable transcription services within their automated workflows and agentic processes.
Fun-ASR Use Cases
- Transcribing multi-format audio files such as meeting recordings or voice memos into text.
- Extracting text content from legacy audio data for indexing and searchability.
- Automating speech-to-text pipelines within larger Openclaw Skills sequences.
- Enabling voice-command interpretation for CLI-based AI agents.
How Fun-ASR Works
- The user provides an audio file or triggers a transcription request through the Openclaw Skills interface.
- The skill validates the file format, supporting extensions like .wav, .mp3, .m4a, .flac, and .ogg.
- The audio data is processed and sent to the Alibaba DashScope API using the provided API credentials.
- The FunASR model performs end-to-end recognition, utilizing its deep contextual training to generate a text transcript.
- The resulting text is returned to the agent or user for further processing or storage.
Fun-ASR Setup
To utilize Fun-ASR within your Openclaw Skills environment, follow these steps:
- Set Environment Variables: Obtain your API key from Alibaba Cloud and export it:
export DASHSCOPE_API_KEY="your_sk_key_here"
- Install Dependencies: Install the required Python libraries:
pip install dashscope librosa
- Execution: Run the transcription script via the CLI:
python scripts/cli.py your_audio_file.wav
Fun-ASR Data Schema & Taxonomy
The Fun-ASR skill manages data through specific input requirements and outputs structured transcription metadata:
| Feature | Specification |
|---|---|
| Supported Formats | .wav, .mp3, .m4a, .flac, .ogg |
| API Provider | Alibaba DashScope (FunASR Model) |
| Output Data | Plain text transcriptions and recognition confidence |
| Versioning | 1.0.0 (Released 2026-03-16) |
Fun-ASR Advanced Features
- Multi-format support ensuring compatibility across various recording devices and software.
- High-accuracy contextual recognition powered by DAMO Academy's large-scale training data.
- Seamless integration with the broader Openclaw Skills ecosystem for automated text analysis post-transcription.
- Efficient CLI interface for headless server environments and DevOps pipelines.
SKILL.md
Loading
Related Openclaw Skills
Local Skill Manager
A centralized management utility for organizing, versioning, and scaffolding AI agent skills within a local directory.
Game Numeric Design
A specialized game design toolkit for creating balanced numeric systems, economic models, and progression curves.
GitHub Repo Quickstart
A developer-first skill designed to distill complex GitHub repositories into actionable, low-friction onboarding guides.
OpenClaw CoE (Chain of Execution)
A native execution chain skill that provides real-time, step-by-step visibility into how OpenClaw processes complex tasks.
News Brief - 中文新闻简报
A local scraping tool for aggregating Chinese news portal content into clean Markdown digests or RSS feeds without requiring external API keys.
Experience Distiller
Experience Distiller intelligently routes completed work into specific knowledge layers to ensure long-term project intelligence and reusability.
Featured*
